So I'll repeat:
1) In a BTL system if one bridge fails you can switch off (tristate) the other bridge. This renders a blown FET totally irrelevant.
2) You also have the option to switch the other side of the failed bridge on - creating a rail-to-rail short to make the device safe.
Actually there is, because you can drive Class D with digital logic gates than can be used for protection.
You can't easily do that with Class A/B, look at the internal circuit diagrams of those chip-amps, analog drive doesn't work very well through logic gates.
ClassD chips have scope for a _far greater_ control over the output devices, which allows many more protection strategies. In this respect (I.e. the thread title) they differ significantly from other classes.
It's a good example though, which is why I used it. I don't really think it's the only class D, I even have a TDA7498 and TDA7492 IIRC here too 😉. Doh!
So to answer the original question of the thread, Class-D chips can have significantly better protection built right in than a different class of amp, although some class A-B chip amps have some protection.
This is an interesting bit of news, what happened?
I think it's always wise to think about fuses (for DC faults) or something like this:
Digital Amplifier Speaker Protection Board BTL for TDA7492 TDA7498 TPA3116 | eBay
if you have something too valuable to lose on anything without an output transformer (some solid state have OPTs too).
The global NFB in a class AB amplifier will correct the output in the opposite direction of the shorted output device, turning on the opposite device and shorting the power supply. Pretty much any class of amplifier will do this, class D included.
This isn't an elegant solution and it's not guaranteed to work, you'll either blow a power supply secondary fuse (assuming it exists), blow the opposite FET taking down the power supply, or blow up something else like a current sensing resistor which could then result in the output being driven to one rail without any means of fixing it.
You don't need a logic gate to turn off an output device. In the analog domain you can turn off an output device by pulling away the base current or gate voltage by any number of different means.
Plus, look up NXP UM10155, which is the very common UcD amplifier design. nCore is the same essential design with a more sophisticated control loop. No logic gates. The D in class D doesn't mean digital.
The TPA311x chips require a minimum amount of inductance on the OUTxx pins to protect against an output short. So if you get a solder bridge between the OUTxx pins and one of the power supply rails at the chip itself, and power the device on, you'll blow output devices.
I powered up one of my Wiener cards using a resistive load, noted the protection LED was blinking, found a solder bridge and put the amp aside to rework the TPA, moving onto the next board. A few days later when the TPA was reworked and the solder bridges cleared, I hooked up speakers and applied power to the thing... CRACK, the cone jumped. By the time I got the power supply disconnected there was smoke coming out of the speaker already.
Thankfully I use cheap car audio speakers bought at a thrift store for testing, so I wasn't out any serious money or anything.
Speaker protection relays usually exist in class AB amps because when you power them on, lots of different parts in their guts in them have to charge up, and this can create a big turn-on thump. The simplest solution to this is to just have a turn-on timer that closes the relay after a pre-set delay, but adding a DC detect circuit that'll also open the relay doesn't require much in the way of extra work/parts and is another check box to put on the amplifier feature list.
True, if there is strong GNFB this would do something.
Agreed, but the drive to the output MOSFETs is however digital. This allows the use of cheap logic gates.
Ouch! Worth adding external protection to then.
Check out the link above - I don't know if they work of course by I'll get one and see what it does under some test conditions.
Size zero, be sexy, be rich. Get respect from the beautifulpeople.
to get rid of the old fashioned large protections.
Beautiful girls you mean ??😀😀
Give him a large chunk of Toblerone to make him understand your point
It was more a side kick. In enjoy to play with words and analogies.
And sometimes I am struggling to stay serious.
Class D amps are the 'beautifulpeople' of the amp world.
There is a rapidly growing trend, especially in the large cities, to judge people by their shape, money and arrogance.
Like some people accept & do everything in order to become beautiful, some other people (but not necessarily other) accept & do everything to make an amp small.
...yes, I am enjoying the world of classD amps since years - now it's up to you to guess about analogies and my life style...
Sorry for OT - have to stop this. Most likely it's already close to the limits of the forum rules.
1. Transformerless tied
2. Capacitor less tied
For sure you need DC Protection
Why make it so complicated? 🙁
2. Capacitor less tied
For sure you need DC Protection
Why make it so complicated? 🙁
£3 for components for a speaker DC protect circuit is well worth the money.
An externally hosted image should be here but it was not working when we last tested it.
My PIC micro based version is totally soft, I can set it where ever I want.
It currently looks for DC for 500mS, if it sees it turns off the relay.
Also on power up there is a 3 second anti thump delay on the relay.
There is also an LED to show when circuit is in protect mode.
There is very little software code to do it.
I have a test bed for new amplifier designs, this PIC circuit has saved me hundreds if not thousands of pounds in failed speakers.
I occasionally make a build mistake and the amp outputs DC.
Good for you Nigel, because you can play with MC.
But I love simple discrete, always works and it is combined with bias supply. And I cant use MC also. 😀
But I love simple discrete, always works and it is combined with bias supply. And I cant use MC also. 😀
Actually you do not need the minimum inductance for the short circuit protection to work. The TPA311x series of chips are designed to work filter free with s simply ferrite bead. The short circuit protection still works even in such a condition.
You're only going to get a solder bridge when your assembling the PCB too.
Which is exactly why you used a cheap car speaker.
No one connects their freshly made amplifier to a pair of expensive loudspeakers without testing it before hand. And even if an individuals test equipment is lacking a digital multimeter to test the outputs is available to everyone, as is a cheap loudspeaker.
I am actually surprised that you managed to damage a TPA311x in that way but '**** happens' as they say.
Considering the thread title though I thought we were actually discussing whether or not class D amplifiers need speaker protection when they are operating properly to begin with. Certainly one could say you need as much protection as possible when a design is powered up for the first time, but that's considerably different to when your amp is working properly.
With TIs range of single chip class D amplifiers, they do indeed have a very extensive range of protection features built in.
The TAS5630 and now TPA325x range even include Pin-to-Pin Short Circuit Protection (PPSC), on top of over current, over/under voltage, after filter short circuit and DC offset protection.
Once such a device is operating correctly in a proven design and has passed the standard batch of initial testing, such as in a consumer product, the question I believe now becomes - Is any kind of protection scheme required beyond what the chip includes on-board?
TI would no doubt say that the on board protection is more than enough and quite frankly I am sure if a DIYaudio memeber had designed a protection system as extensive as TIs on-board one and had implemented it in their discrete design that they'd think it was more than enough too!
Bottom of page 5, minimum inductance 1uH:
http://www.ti.com/lit/ds/symlink/tpa3118d2.pdf
Believe me, I can find a way to break anything.
Car speakers are fabulous test loads for the price. The JVC ones I'm using right now came from a thrift store for $5, and the previous Pyramid set which I "half blew up" with the failed Wiener card cost a whopping $10 I think.
I think for any professional amplifier design, having output relays and DC detect/disconnection an is essential feature.
Pro audio speakers (like Peavey/JBL/etc) aren't cheap. And professional amps get carried around a lot and get dropped/beaten up, miswired, get who knows what spilled into them... there's all kinds of opportunities to make them fail. And when that failure inevitably happens, it should not take out the attached speaker with it.
Whatever is built into the class D chip itself might be completely bypassed if there's a dent in the chassis touching the back of a PCB 🙂
Agreed, the dual cone (whizzer cone) ones also sound pretty good too in my experience, a lot better then the cheap Eltaxes I use as test speakers!
With all that on-chip protection of the TI range I suspect a simple fuse would be good enough for any faults that 'get away'.
Much of the old Japanese amps use them, 1.5 or 2.5A rating long fuses as I recall.
Jumping into the thread late here, but class-D amplifiers are, in their basic form, fully analog.
Google Bruno Putzeys' original article on self-oscillating class-d amplifiers, and look at his circuit. There's an analog comparator (a differential amplifier driven to saturation) to turn on the gates, but no digital logic. More traditional designs would use a triangle waveform generator instead of self-oscillating, but the principle remains the same: the drive signal is an analog square wave.
The output transistors are either 'on' or 'off' are they not?
That doesn't look fully analog to me, I may be old fashioned but to me it seems to be binary, or 'digital'. A 1 or a 0.
As for their 'basic form' - what does that mean? Do we use those types? For the chip based class D, what proof have you that TI etc. use no digital logic between the PWM generator and the gates? I'd love to see the schematics, because the block diagram shows 'PWM logic' and 'Gate drive' blocks. Is that analog logic?
In many ways, the Intel i7 CPU could be described as 'fully analog' because it's not a quantum computer and it thus it has transistors driving other transistors, with rise times etc just like in class D. So at what stage do we decide the names 'digital' and 'fully analog' diverge?
Surely at the end of the day, all electronics is fully analog?
I always put a scope on the output of new amp before powering it up.
If the output goes low or high I turn it off straight away.
I also just use a 2 amp power supply for first test, the mosfets will survive 2 amps most times if there is a problem.
I also have a speaker DC protection box on the output of the amp while I burn them in.
Having spent my life repairing practically every mode of output stage failure, many of which resulted in speaker destruction or damage, I would say there is absolutely no reasonable reason to not design comprehensive speaker protection into either the amplifiers you build or alternatively, implement a self powered speaker protector in-line with your priceless speakers.
Just because some commercial, audiophile Class A, seat-of-the-pants designs eschew protection, means nothing to me. The designers/marketers don't care about your speakers. If I had an expensive commercial design fail and take out some ludicrously priced speakers, I'd be taking on the manufacturer for consequential loss. All electronic equipment should be designed to be safe. Safe in operation and not cause potential fires or destruction of attached components when unforeseen circumstances arise . I've seen speakers completely burnt out and needing a fire extinguisher due to DC failures of their upstream amplifiers.
Incorporate the tried and true turn-on delay, over current and DC in an amplifier or over current/voltage and DC if you build it into the speaker line. Forget messing around with PICs and code for a speaker protector- they are not robust or reliable enough. What's wrong with a uPC1237? Here's a link to a sweet little project (no-afil)
uPC1237 Amplifier/Speaker Protection Module - DIY AUDIO BLOG, AUDIO WORKSHOP
just do a search on that auction site for upc1237 based protectors- they are as cheap as chips.
Just because some commercial, audiophile Class A, seat-of-the-pants designs eschew protection, means nothing to me. The designers/marketers don't care about your speakers. If I had an expensive commercial design fail and take out some ludicrously priced speakers, I'd be taking on the manufacturer for consequential loss. All electronic equipment should be designed to be safe. Safe in operation and not cause potential fires or destruction of attached components when unforeseen circumstances arise . I've seen speakers completely burnt out and needing a fire extinguisher due to DC failures of their upstream amplifiers.
Incorporate the tried and true turn-on delay, over current and DC in an amplifier or over current/voltage and DC if you build it into the speaker line. Forget messing around with PICs and code for a speaker protector- they are not robust or reliable enough. What's wrong with a uPC1237? Here's a link to a sweet little project (no-afil)
uPC1237 Amplifier/Speaker Protection Module - DIY AUDIO BLOG, AUDIO WORKSHOP
just do a search on that auction site for upc1237 based protectors- they are as cheap as chips.
Last edited:
PIC's are very cheap, I think I paid about 60p for the one I use in a speaker protector circuit. The code for speaker protection is just a few lines.
PIC's are also very hardy and reliable. I have seen many people poke mains into PIC i/o ports through a high value resistor.
They PIC's also have a watchdog circuit and brown out detect.
Yes, I agree they are cheap.
That's where we perhaps disagree. In my limited experience with the AT Mega chips, they are easily taken out through the I/O ports when simply using high value resistors. I have several here I killed with circuits where the resistors should have been sufficient for the applied voltage ranges I was working with. The interesting thing with the AT Megas is their failure mode- they basically have gone internal short on the I/O lines, pulling 300mA+ now they are dead. I would think opto-coupling would be the only safe front end for them now. High powered amplifiers can generate phenomenal transients and spikes in many fault/overdrive conditions and I don't believe a resistor coupled PIC protector would be reliable in the long term.
The 30 year old UPC1237 has AC detect, overload and DC all on a single rail SIL chip.
PICs are fun, I'll give you that, but why complicate a speaker protector by adding a computer? By all means if it is a full blown amplifier protector and system monitor- go crazy, but speakers really only care about too much DC, too much power or nasty thumps at turn on or off.
The same rules apply to any class of amplifier, and with these 'class D' amps being acting like noisy oscillators most of the time, protection is a must IMO.
Speakers should have a sticker that says "Don't turn me on without adequate protection" 🙂
- Status
- Not open for further replies.